Cognitive Simulation of Qualitative Symmetry Detection

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

The objective of this research is to develop a self-adapting vision system inspired by biological systems. The proposed system will construct models of its environment and will employ feedback processes in a hierarchical framework based on natural systems. The project will investigate a new class of symmetry phenomena involving approximate or qualitative symmetry; qualitative symmetry is beyond the scope of most cognitive models of perception and humans often classify nearly symmetric figures as not symmetric. The model offers the potential of explaining a number of unintuitive and surprising aspects of human performance on the task. The work proposed is aimed at fleshing out and testing these explanations, and secondarily at developing a robust modeling tool for this task. The proposed research is meritorious intellectually since it addresses the problem of qualitative symmetry detection, an aspect of qualitative spatial reasoning, a topic that is a very important in artificial intelligence. This work has a broad impact on many areas important to computer science and engineering including computer-aided design, interpretation of diagrams and the creation of diagramatic representations. This work will also have a broad impact on understanding human qualitative reasoning in general.